On Noise, Coax, and Control…

By Bryan Geyer

In an earlier paper, I wrote about all of the various kinds of cordage that gets utilized when interconnecting the components that comprise home stereo systems. (See “On Equipment Interface Options”, at https://classicalcandor.blogspot.com/2019/10/on-equipment-interface-options.html.) I’m now going to discuss a recent development that involves only the most popular of these various interconnecting cords, the kind of shielded coaxial cable (optimally type RG59U, characteristic impedance 75Ω) that’s commonly mated to RCA-type plugs and used to interconnect unbalanced inputs and outputs.

The EMI (electromagnetic interference) environs in a private home are not like those at a rock concert venue. In most homes, noxious EMI and RFI noise is minimal. There’s no strobed lighting, no motorized generators, no high output DC-to-AC inverters, and no RF transmitters. (But check the premises for wall-warts. Some old switching supplies might impose an illegal [not code compliant] EMI threat when in use.) Further, the cable runs required in a home installation are fairly short, and without the need to accommodate frequent disconnects and reconnects (“hop ons” and “hop offs”). When this benign reality applies, the noise implicit with unbalanced interconnects will be just about the same as that apparent with balanced cables. In such case, the much lower cost, inherent simplicity, easier handling, and space-saving advantage (RCA jacks consume less than half the chassis space as that required for XLRs) of RCA-type terminations make unbalanced cables the logical preferred choice. I personally encourage the use of unbalanced cables, preferably using RG59U coax, for most home stereo installations. Consider balanced XLR cables when some pending change in the site environs threatens to significantly increase the prevailing noise.

A NOISE TIP: Turn all of your audio components on, and select an unused input (or select your CD player without loading a CD). Make certain that your power amplifier is active; turn it on manually if it’s asleep and normally activated only by sound. Turn the main volume control full up, and press your ear against the grill of either main speaker. Can you detect a faint buzz? If so, that background noise can often be eliminated by running an 18 gauge wire from a signal ground (use the outer shell of any RCA input jack that’s on the back panel of your active or passive preamp) directly to the AC line ground port of the alternate AC socket that’s on the same duplex outlet as the one used to connect your main AC line strip outlet. Check this… https://www.statictek.com/product/ground-plug-adapter/ …for a convenient way to make that simple connection by means of a banana plug adapter.

The best RG59U (75Ω) coaxial cable that you can buy is made by Belden; it’s their type 1505F, and it’s stocked by Blue Jeans Cable. I prefer this coax to Belden’s other premium coax cables that have a soft-foamed dielectric layer because 1505F is compatible with soldered-on RCA plugs. The soft-foamed coax equivalents provide slightly less shunt capacity, but you’re then confined to the exclusive use of crimp-on type RCA plugs, e.g., plugs from Canare and Taversoe. Crimp-type plugs are unusually long; they consume too much rear clearance. With Belden 1505F, you can order Rean’s soldered-on RCA plugs. I find the Rean (they’re a part of Neutrik) soldered-on plugs to be more rugged and less fussy—and they take less clearance than the longer crimp-style RCA plugs (see photo).

Belden 1505F cable exhibits a shunt capacity rating that’s still commendably low, at just 17pf/foot. Other unbalanced audio interconnect cable ranges from 12.2pf/ft. (for the softer air-foam coax) up to some 35+ pf/ft. With respect to shielding, Belden 1505F is almost as good as it gets. It has a densely woven double layer of copper braiding, one layer atop the other, with each layer providing a calculated 94% coverage. The PVC outer jacket Ø is 0.242 inch, slighter slimmer than the soft-foam coax cords at Ø ≈ 0.305 inch.

To order Belden 1505F coax with soldered Rean RCA plugs from Blue Jeans Cable, go to…https://www.bluejeanscable.com/store/audio/index.htm, and scroll down to the next-to-last option box, where the heading reads “Belden 1505F Stereo Audio Cables”. Fill in the precise length(s) that you want to order (expressed in feet); note that the price is per pair (currently $42.25 for a 3 foot long pair). Order whatever lengths and quantities that you want. Forget the Techflex. Put your order in the shopping cart and proceed with the ordering process. During checkout, on the last page that gives you your shipping options and final pricing, you will see a text box to permit leaving a note. Use that text box to instruct Blue Jeans that you want them to…

Use Rean solder-on RCA plugs with 1505F. Do NOT use Canare crimp-on plugs.

If you fail to enter this instruction in the text box you will receive Canare crimp-on RCA plugs on your cable.

Blue Jeans further advises that you should not use the PayPal “quick checkout” option, as that will sometimes bypass their text block page. There is no additional cost for this special processing. The extra expense incurred in their precision soldering task is offset by the lower parts cost of the Rean (instead of Canare) RCA-type plugs.

Be assured that Belden 1505F coax cables with Rean soldered-on RCA plugs are among the finest quality unbalanced interconnect cables that you can buy, at any price. Assuming a direct aural comparison under controlled double-blind test conditions, you will be absolutely unable to distinguish any difference between these cables and any more costly equivalent. This challenge includes direct comparison with such fatuous substitutes as this ultra high-end product: https://www.synergisticresearch.com/cables/atmosphere-x/atmosphere-x-ic/euphoria-level3/.

A DISCLAIMER: Given the potential tuning tweaks possible with this functionally equivalent cable, it could readily alter the accuracy (distort) the incoming source signal. In such case, an aural difference might then be apparent. The nature and extent of that distortion would be evident on instrumented measurement. (It’s likely to involve frequency response.) You might (???) prefer the distorted sound, but you’d be better served by selecting the cable that delivers the least change, hence best accuracy. A measurement would reveal the inaccuracies.

REGARDING CABLE SHUNT CAPACITY: The importance of minimizing the shunt capacity (hence the length) of unbalanced interconnecting cables will vary, dependent primarily on the value of the signal’s source impedance. The focus of concern is whether the net shunt capacity of the cable is enough to potentially degrade high frequency linearity; i.e., roll off the treble response.

In the event that your system involves a conventional solid-state setup, with various line-level sources feeding into an active solid-state preamp, and that preamp then connected to a stereo power amplifier (or to an external electronic crossover controller) with a high input impedance (i.e., Zin ≥ 30kΩ), you can dismiss concern about excessive cable length. In those cases, the output impedance of your source will always be quite low (it’s feeding the next stage from an emitter follower, and Zout is ≤ 100Ω), so you really won’t have to worry about the cables getting too long. If you’re still mired in the vacuum tube era, your source impedance from a cathode-follower stage will characteristically be some 6X to 10X worse (Zout ~ 450 to 700Ω) than from an emitter-follower, so yes, do be a bit more conscious of cable length concern. Assuming common 120pf-per-meter cabling and a vacuum tube cathode-follower stage with Zout = 600Ω, the cable length to the next high impedance load (i.e., the power amp) should not exceed ~ 15 meters. Of course, half of that length is enough for almost any sensible home installation.

However, the situation can change appreciably when you employ a “passive preamp”. (A passive source-selector box, where the output signal is taken directly from the wiper of the volume control attenuator.) In that case, the attenuator’s worst case (highest) Zout will be considerably > the Zout from an active emitter or cathode follower, and that increased Zout will materially impact permissible cable length. Examples: (a) If the attenuator was a log taper 50kΩ potentiometer (pot), it’s worst case Zout would be ~ 12.5kΩ, so you’re likely to initiate some treble roll-off (assuming 120pf/meter cable) if the length was > 1 meter max. (b) If using a 25kΩ pot, the limit would be 2 meters max. (c) If using a 10kΩ pot, the limit would be 4 meters max. In the event that you use coax that exhibits a shunt capacity > 120pf/meter (yes, it’s out there), your cable lengths should then be even shorter.

Do also bear in mind that these calculations assume that the ensuing stereo power amplifier (or external active crossover controller) presents a relatively high load impedance; something on the order of 30kΩ to 50kΩ or more. While this is generally the case, one prominent producer of hi-end stereo power amplifiers offers a model that exhibits a Zin of only 10kΩ (unbalanced), or 15kΩ (balanced); refer…https://www.anthemav.com/products-current/type=amplifier/model=mca-225/page=specs. That unusually low Zin would negatively impact these cable length calculations, and it would make this product a poor match for use with any passive preamp, although tolerable if the attenuator was 10kΩ. I am not aware of any other commercial audio power amplifier on the market that exhibits such extremely low input impedance as (some of) this company’s power amplifiers.

THE BUFFER OPTION: You can completely avoid all concern about the consequence of loading effects (even when it’s as low as 10kΩ) by inserting a unity gain buffer (UGB) stage. The UGB simulates an inviolate brick wall. It isolates the passive preamp’s volume control from all external loading other than that imposed by the UGB itself. A UGB is especially desirable in the event that you intend to use a premium volume control with precisely calibrated incremental (stepped) attenuation. The UGB should then be designed to present a fixed high impedance, something on the order of ~ 15X ≥ the worst case (highest) Zout of the attenuator; e.g., Zin ≥ 75kΩ in the case of a 20kΩ calibrated attenuator. (A 20kΩ pot’s worst case Zout = 5kΩ. The “worst case” always equates to the pot’s -6dB down point.) This would assure an absolute maximum loading error of -0.5dB at the -6dB position of the attenuator, with progressively less loading error at all other positions of that calibrated control. The UGB stage should further provide: Low Zout (~ 50Ω), minimum distortion, optimum linearity, and it should utilize high value input and output coupling capacitors (preferably of non-polar polypropylene) for good low frequency response.

Top quality UGBs are inherently simple in design. They’re often configured as discrete NPN/PNP complementary feedback pairs (CFPs)*, optimally with constant current biasing. CFP design is thoroughly addressed in Douglas Self’s book Small Signal Audio Design. (See new 4/22/2020 3rd edition, at…https://www.routledge.com/Small-Signal-Audio-Design/Self/p/book/9780367468958.) High performance UGBs of this sort will assure that the calibrated precision of the attenuator remains essentially as designed, free of the effects of any external loading that’s beyond the buffer. THD will be ≤ 0.001% for output swings ≤ 5Vrms, and noise will be inaudible when the UGB is constructed to observe star grounding, and fed from a separate linear regulated supply. I designed my own dual-channel UGBs to run off of a single-ended linear +40Vdc regulated power supply, using Acopian’s 40EB06 miniature module…https://www.acopian.com/store/productdetail.aspx?q=dNominal+Output+Voltage:+40;+Output+Current+Amps:+0.06;+++Regulation+Load+%2B%2F%2D:+0.02;+Line+%2B%2F%2D:0.02;++Ripple+mV+RMS:+1;+Case+Size:+EB-13,i674. My dual channel CFP UGBs draw ~ 36mA net; the Acopian 40EB06 is rated 60mA max.

Audio hobbyists have traditionally utilized a standard (active) preamplifier box when configuring an audio system. That habit stems from the days when vinyl was the principle hi-fi signal source. Today, anybody that’s still playing vinyl usually locates their low-level gain stage (with +40dB [MM] or +60dB [MC] of RIAA compensated boost) separately, nearer the phono cartridge, so all of the source signals now routed to a traditional preamplifier box are already at full “line level” amplitude; i.e., a level of some 2Vrms or more. In such event, there is then no need for any further amplification. The available line level signal is already sufficient to drive the stereo power amplifier to the full extent that your audio system can output.† Indeed, further line level gain is distinctly undesirable, as it would then force the volume control to be operated at unduly excessive loss.

A traditional active preamplifier box retains two lingering benefits: (1) It enriches the parties that make and market those relics, and (2) it normally provides a basic emitter follower output stage (a simplified UGB equivalent) to isolate the internal volume control potentiometer against the effects of external loading. You’d do well to…

            (a) ditch the preamp.

            (b) substitute a box with a superior (accurately calibrated) stepped volume control attenuator.

            (c) include a basic signal source selector switch inside your new box.

            (d) consider adding a capable UGB/CFP stage. (Much better performance than a simple emitter-follower.)

Steps (b) and (c) of this sequence can be accomplished as either a DIY effort (in which case you’d assemble your own calibrated attenuator, using multiple discrete ±0.1% tolerance metal film resistors), or by purchasing a commercial “passive preamp”, such as those offered by Goldpoint Level Controls (https://goldpt.com/index.html).

Stacking a UGB stage onto the output of a volume attenuator is mostly a matter of personal preference. The benefit hinges on (a) the value of the attenuator that you’ve chosen, (b) the physical distance to the next high impedance load (your stereo power amplifier or external active crossover controller), (c) the actual value of the Zin at that input, and (d) the shunt capacity rating of the connecting cable. The UGB simply assures that the elegant precision of the calibrated attenuator won’t be sullied by any of those external influences, regardless of how they might vary, now or in the future. As a result, the true advantage conferred by adding a UGB might not amount to anything more than perfectionist’s pride. Of course, the aural perception of such pride would admittedly be quite subdued, although it’s likely to be audible to anyone who’s ever added a DIY UGB to their system.

PERSONAL REPORT: I have been using a passive preamp ever since 1984, and I’m now building my third generation version. Both of my earlier passive preamps performed well—I just lust to update, and three times in 36 years seems reasonable. My latest DIY effort features a 20kΩ stepped attenuator that’s scaled at -2dB/step over an initial spread of 17 steps (-34db), tapering to a total of -62dB over 6 additional steps prior to reaching the final fully-off (grounded) position. I use discrete 1/4 Watt low noise metal film resistors with ±0.1% (repeat, ±0.1%) tolerance (sourced from Mouser), and mate them to a premium grade Goldpoint model V24C-2 (Elma type 04) double-deck rotary switch with 24 stepped positions.

I’m concurrently building my fourth DIY unity gain buffer. While my previous UGBs were entirely satisfactory (yes, they all sounded the same), my latest CFP (complementary feedback pair) circuit is of a more esoteric design. In addition, this new buffer will utilize polypropylene (rather than PET film) non-polarized coupling capacitors.

DIY audio engineering can be a rewarding endeavor. It’s often enlightening, sometimes humbling, and always instructive. The most gratifying DIY advantage is that your project can always be better than the best that you can buy. The bounds of commercial practicality don’t apply when romping in your own DIY playpen.

BG (August 8, 2020)

*An IC op amp, configured as a simple voltage follower, can be used as a buffer, but the related supply voltage limitation would then restrict the output swing to relatively modest peak amplitudes. A better UGB can be built by using discrete transistors, configured as a CFP, and operated from a higher voltage supply; e.g., ±20Vdc as a split supply, or +40Vdc as a single-ended supply. This will assure that peak signal swings are well within ultra-linear areas of the waveform. The output drive capability of the CFP will also be superior to that of any IC op amp. 

†This is most certainly the case when the power amplifier presents a voltage gain ≥ 26dB (≥ 20X). It’s also highly probable—if not absolutely certain—that power amplifiers with a voltage gain as low as ~ 23dB (14.1X) will also prove fully compatible. The voltage gain of a power amplifier is commonly stated in the product specifications. If not so stated, voltage gain can then be derived from the specification for input sensitivity.

ADDENDA #1: Do appreciate that the voltage gain parameter (cited above) is merely a measure of an amplifier’s sensitivity. It has no direct bearing on defining the power output capability or power output rating of an amplifier.

ADDENDA #2: The actual power output capability of the main power amplifier should be at least +2dB (1.6X) to +3dB (2X) > the rated power that the loudspeaker system can safely tolerate. This will assure that the loudspeakers are never exposed to a clipped input signal when they’re driven to levels that are within their maximum safe rated limits. If your main power amplifier has accurate “clipped output” warning lights that sometimes flicker, that amplifier has insufficient power output capability for its application; it should be replaced. Outside of physical abuse, nothing can be more potentially injurious to a loudspeaker system than consistently clipped drive signals, and nothing sounds worse than peak level clipping.